Re: [PATCH v5 02/15] read-cache: allow to keep mmap'd memory after reading

From: David Turner <dturner@twopensource.com>
Date: Apr 20, 2016, 19:43 UTC
Message-ID: <1461181395.5540.170.camel@twopensource.com>
In-Reply-To: <CACsJy8D229ARqEjH29Gqd=u8KijGA_Fzk0cbUNRXB9GA0ittMg@mail.gmail.com>

On Wed, 2016-04-20 at 16:26 +0700, Duy Nguyen wrote:

Show 18 quoted lines

> On Wed, Apr 20, 2016 at 6:27 AM, David Turner <
> dturner@twopensource.com> wrote:
> > From: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> > 
> > Later, we will introduce git index-helper to share this memory with
> > other git processes.
> > 
> > Since the memory will be shared, it will never be unmapped
> > (although
> > the kernel may of course choose to page it out).
> 
> This is not entirely true. We do need to keep the mmap'd memory for
> longer, even after read_index_from() finishes. But we do not share
> the
> exact same address space to other processes (memcpy is used in
> index-helper.c:share_index()). We could even discard_index() at the
> end of share_index(), but I think we keep it anyway just in case
> another process asks, or when index is not updated.

Will reword.